Premier League predictions using xG

best odds using OddsChecker

Liverpool 2-0 Watford

9.00

Burnley 1-0 Newcastle

7.00

Chelsea  2-0  Bournemouth

8.50

Leicester     2-1 Norwich

11.00

Sheff Utd    2-1  Aston Villa

9.50

Southampton    1-1  West Ham

8.50

Manchester United  2-1  Everton

9.00

Wolverhampton Wanderers    1-1  Tottenham Hotspur

7.50

Arsenal  1-2  Manchester City

10.50

Crystal Palace  1-1   Brighton & Hove Albion

7.00

Premier League predictions - week 17
....Result    ...Lawro    ....David baddiel ......you have to laugh
SATURDAY           
Liverpool v Watford    x-x    3-0    4-1    3-0
Burnley v Newcastle    x-x    2-0    2-2    1-1
Chelsea v Bournemouth    x-x    2-0    3-0    4-1
Leicester v Norwich    x-x    3-0    3-1    3-0
Sheff Utd v Aston Villa    x-x    2-1    1-0    2-0
Southampton v West Ham    x-x    2-1    2-2    2-3
SUNDAY           
Man Utd v Everton    x-x    2-0    2-1    3-1
Wolves v Tottenham    x-x    0-2    1-2    1-2
Arsenal v Man City    x-x    0-2    2-3    3-3
MONDAY           
Crystal Palace v Brighton    x-x    2-1    1-0   1-1

^^^^cut and paste from bbc, with mine tagged on end

good luck betting masta

good start masta

looking decent if chelsea can open things up and if leicester can grab a winner

by all accounts bournemouth just had a great chance to score

CLOSE!
Chelsea 0-0 Bournemouth

What a chance for Bournemouth!

Joshua King manages to get in behind the defence. He thinks about a shot on goal but the angle is tight so instead flicks it across for Dan Gosling, who looks certain to stab into the empty net but misses his shot.

Lucky, lucky escape for Chelsea.

Liverpool 2-0 Watford

9.00 WON

Burnley 1-0 Newcastle

7.00 WON

Chelsea  2-0  Bournemouth

8.50

Leicester     2-1 Norwich

11.00

Sheff Utd    2-1  Aston Villa

9.50


--
two others were close, villa missed a late pen for 2-1, leicester could have nicked it

yeah, good stuff.

well done masta

very close again , 5 of the last 6 games out by the odd goal

3 right, good stuff, i got 3 too, easy game ...lol

Top tipping

Decent effort Masta

Harder than it looks.

Average number of goals per game over the last few seasons averaging 2.7 to 2.8 so you're pitching it about right.

Getting the score right is another matter

cheers guys

+13 profit after Round 1 - i will continue for a few weeks to see if there's something in the specific method, it's merely using the data and so i take no credit, it's just the data being interpreted and playing out

the odds suggest we should be getting 1-1.5 correct score per round so 3 is above average, we'll see if it is just "luck" - probably ;)

this xmas period could see some variance due to team changes etc but should even itself out

back on friday for round 2

Excellent start bm; I've been tracking Premiership correct scores at £10 per game (on paper) and after 5 weeks I'm £9.20 down via Exchange prices (allowing for commission) backing just what I 'feel' will be the correct score on each game. Will be good to see this one develop. Good luck!

forgot about this for a few days --- picking it up again -- exchange price in brackets

Brighton 2-1 Bournemouth
(9.60)

Newcastle 1-1 Everton
(7.20)

Soton 1-1 Palace
(7.80)

Watford 2-1 Villa
(9.80)

Norwich 1-1 Spurs
(9.00)

West Ham 1-2 Leicester
(9.80)

Burnley 1-1 Man Utd
(8.00)

Arsenal 1-1 Chelsea
(8.80)

Liverpool 2-1 Wolves
(10.00)

Man City 3-1 Sheff Utd
(12.00)

average goals per game 2.60 -- about right -- let's see how it does

2 correct again
+7.6

+20.6 after 2 rounds

the experiment continues. i've formulated the calculations now, nothing complicated, should be even more accurate, just doing what the data says will happen
taking exchange price for most value

Brighton 1-1 Chelsea
8.40

Burnley 2-1 Aston Villa
9.40

Newcastle 1-2 Leicester
9.60

Soton 1-1 Tottenham
8.40

Watford 1-1 Wolves
7.20

Man City 2-1 Everton
10.50

Norwich 1-1 Palace
7.20

West Ham 1-1 Bournemouth
8.20

Arsenal 1-2 Man Utd
10.50

Liverpool 2-1 Sheff Utd
9.80

-good luck-

projected avg goals per game is 2.50-- that's a bit lower than the average for the season in the premier league which is 2.81 so that tells me the model is slightly undershooting due to the rounding down. i've essentially taken the average for xG for the respective teams home/away, if anyone has any ideas to make the figure more accurate than please do tell, i'm pretty much a novice at excel/mathematics

3 winners this week -- strike rate is healthy  (8 winners out of 30 matches is 26.67% strike rate) a couple of close ones as well

week +16.10
running total +36.7

Sheff United    2-1 West Ham
10.00

Palace    1-1    Arsenal
8.60

Chelsea    2-1    Burnley
10.00

Everton    2-1    Brighton
9.80

Leicester    1-1    Southampton
9.60

Man U    2-1    Norwich
9.80

Wolves    2-1    Newcastle
9.80

Spurs    1-1    Liverpool
9.60

Bournemouth    2-1    Watford
11.50

Aston Villa    1-2    Man City
12.00

Palace    1-1    Arsenal
8.60
WON  +7.60

week -1.4
running total +35.3

Looks like you're doing well. Not sure how true this is but read a stat over the weekend (in comparison to Man City's superior goal difference than Liverpool's) that Man City would be top on xG. That's obviously only the one example but does show some possible fallacy in xG.

Week 23 predictions

Wolves 1-1 Spurs
7.80

Arsenal 1-1 Sheff Utd
8.00

Brighton 2-1 Villa
9.40

Man City 2-1 Palace
15.00

Norwich 1-2 Bournemouth
14.00

Soton 1-1 Wolves
7.20

West Ham 1-2 Everton
11.00

Newcastle 1-2 Chelsea
9.60

Burnley 1-2 Leicester
9.40

Liverpool 2-1 Man United
9.60

Amendment
Watford 2-1 Spurs
14.00

Watford have a very high xG at home (1.63), Spurs away xG is (1.52). I put the wrong data in initially so this comes out as a Watford win

this weekend i will also do a first goalscorer based on Xg in each game. this is based on best xG per 90 minutes played  of all players likely to start. i don't expect this to do as well. FGS market seems a bit more finer margins to me. standout prices in bold

Watford    vs Tottenham
Deeney  7.20

Arsenal vs Sheffield United
McGoldrick 9.60

Brighton    vs    Villa
Maupay 4.10

Man City    vs    Palace
Gabriel Jesus 3.80

Norwich    vs    Bournemouth
Pukki 4.10

Soton    vs    Wolves
Ings 4.40

West Ham    vs    Everton
Calvert-Lewin  5.10

Newcastle    vs    Chelsea
Tammy Abraham 4.10

Burnley    vs    Leicester
Chris Wood   7.20


Liverpool    vs    Man United
Salah 4.90

Deeney pen saved at 0-0 unlucky

Pukki landed

going to knock first goalscorer on the head, don't think it will work and don't have the time to do it

xG predictions
week -2
overall +33.3


week 24

Aston Villa 2-1 Watford
12.50

Bournemouth 1-1 Brighton
7.40

Crystal Palace 1-2 Southampton
12.00

Everton 2-1 Newcastle
9.60

Sheff Utd 1-2 Man City
9.40

Chelsea 2-1 Arsenal
9.40

Leicester 2-1 West Ham
9.60

Spurs 2-1 Norwich
9.40

Man United 2-1 Burnley
9.60

Wolves 1-2 Liverpool
9.60

Hi BM

>>> Watford have a very high xG at home (1.63), Spurs away xG is (1.52).

1.63 for Watford seems OK - but Spurs are now 1.16 [XG Away] (as of 20th Jan)

Spurs could not have dropped that much (from 1.52 as you are saying) - as they only have registered a 0-0 draw against Watford
The 1.52 must been for XG-AGAINST (i.e. goals scored against SPURS)

^^yep that's what i meant, xGA. wasn't clear in my message

Aston Villa 2-1 Watford
12.50

+11.50

profit then already  but could have been more

Chelsea 2-1 Arsenal
9.40

Everton 2-1 Newcastle
9.60

both gubbed in injury-time

villa 2-1 was 96 mins
everton 2-1 was 94 mins
everton 2-2 was 95 mins
chelsea 2-1 84 mins
chelsea 2-2 was 87 mins

an exciting few minutes there

no doubt a few grand gubbed at very short odds....

Aston Villa 2-1 Watford

Spurs 2-1 Norwich

Wolves 1-2 Liverpool

3 correct results

week  +21.5
overall +54.8

6 rounds played = profit of 9.13 per round
ROI is very good atm.  91.3%

next round will be 2 games for liverpool including the west ham fixture and then a full round next weekend

round 7

West Ham 1-2 Liverpool
10.00

predictions
Saturday
Leicester 1-1 Chelsea
8.20

Bournemouth 2-1 Aston Villa   
10.00

Crystal Palace 1-1 Sheffield United   
6.80

Liverpool 2-1 Southampton   
9.80

Newcastle 1-1 Norwich   
7.40

Watford 2-1 Everton   
12.00

West Ham 2-2 Brighton   
15.50

Man United 1-1    Wolves   
7.00

Sunday
Burnley     1-1 Arsenal   
8.00

Tottenham 1-2 Man City   
9.40

==

Going well, but will it work after transfer deadline day?

do brighton have an xg of 2 ?

still going well!

Jan 31, 2020 -- 2:36PM, Crisp77 wrote:

Going well, but will it work after transfer deadline day?

Yes, not many players changed hands as it goes before the noon cut off

Jan 31, 2020 -- 3:08PM, ----you-have-to-laugh--- wrote:

do brighton have an xg of 2 ?still going well!

it's relative to the opposition's defence, west ham's in this case being quite weak, as you saw in the 3-3 game

Feb 1, 2020 -- 9:10AM, Crisp77 wrote:

Jan 31, 2020 --  2:36PM, Crisp77 wrote:Going well, but will it work after transfer deadline day?Yes, not many players changed hands as it goes before the noon cut off

the problem is little nuances like a change of manager means a team may set up differently and these predictions look at the entire season, not just the last 10 games. L10 could be more precise and is something to consider. as mentioned, the transfer window didn't see too much movement and so teams basic set ups are roughly the same. i'd say a change of manager would have as much of an impact

week  +1
overall +53.8

a small loss this weekend but it got a fair few results correct (newcastle, west ham, man united, burnley matches all finished as a draw). the danger is obviously a complete blank week which is possible but hasn't happened yet but we have enough profit in the bank for that not to be completely disasterous

*week -1

had some hard lines again, but i guess thats maybe always the case in
low scoring sports.

keep up the good work!

7 out of 10 results right

Feb 3, 2020 -- 10:08PM, Crisp77 wrote:

7 out of 10 results right

YEP, i haven't been tracking match odds results but i imagine the market is much tighter so the margin for profit is lower so that's why i've gone for the CS market as nobody really bets it

Gameweek 28

Everton    2-1 Crystal Palace
10.00

Brighton 2-1 Watford
10.50

Sheff United 2-1 Bournemouth
10.00

Man City 2-1 West Ham
17.00

Unlucky Everton 88 minute goal

Sheff utd score correct;

projected avg goals per game is 2.50-- that's a bit lower than the average for the season in the premier league which is 2.81 so that tells me the model is slightly undershooting due to the rounding down. i've essentially taken the average for xG for the respective teams home/away, if anyone has any ideas to make the figure more accurate than please do tell, i'm pretty much a novice at excel/mathematics

_____________

if you take Manchester City out of the equation the  games average around 2.65   ...amongst lower half teams playing each other it averages  below 2.45

the top 4 or 5 teams normally skew the average somewhat  ...esp last season

ty for that

=Friday=
Wolves 1-2 Leicester
12.50

=Saturday=
Southampton 1-1 Burnley
8.40

Norwich 1-2 Liverpool
9.80

=Sunday=
Villa 1-2 Spurs
10.00

Arsenal 2-1 Newcastle
no price yet

=Monday=
Chelsea 2-1 Man Utd
no price yet

=Sunday=
Arsenal 2-1 Newcastle
9.60

=Monday=
Chelsea 2-1 Man Utd
9.60

week +0
overall +53.8

next round coming soon

=saturday=
Chelsea    2-1    Tottenham
(9.40)

Crystal Palace    1-1    Newcastle   
(7.40)

Burnley    2-1    Bournemouth   
(10.50)

Southampton    2-1 Aston Villa   
(9.40)

Sheff United    2-1    Brighton   
(10.00)

Leicester    1-2    Man City   
(9.60)

=sunday=
Man United    2-1    Watford   
(9.80)

Wolves    2-1 Norwich   
(9.60)

Arsenal    1-1 Everton   
(8.20)

=monday=
Liverpool    2-1 West Ham   
(11.50)

great start vwd

loss of 0.6 units last week
new total +53.2

=Friday=
Norwich 1-2 Leicester
9.80

Norwich    1-2    Leicester   
9.80

Brighton    2-1 Crystal Palace   
10.50

Bournemouth    1-2 Chelsea   
9.40

Everton    1-1 Man United   
7.20

Newcastle    1-1 Burnley   
6.80

Tottenham    1-1 Wolves   
7.00

West Ham    1-2 Southampton   
11.00

Watford    1-2 Liverpool   
9.20

Brighton look a good bet at 2.20 based on xG as they have the biggest xG differential of all home teams playing this week (0.595) - Brighton's home xG 1.400 is pretty good and Palace's xGa (xG against) is 1.81 giving Brighton an xG of 1.605 for this game

Leicester (0.620) was the biggest away differential this week but we saw them lose at Norwich. Vardy being out could have a bit to play in that

Liverpool are the next best away differential (0.450) this week but they're odds-on away at Watford

The game with the lowest xG is, unsurprisingly, Newcastle vs Burnley

Highest combined xG is, maybe surprisingly, West Ham vs Southampton - these are two of the teams with the highest xG and xGa over the season so we can expect over 2.5 goals there

games going to form at HT, newcastle burnley locked at 0-0 and west ham soton 2-1 and looking like further goals

small loss again
bit lucky with that everton united game ending 1-1 as there were big chances near the end

-0.8
new total +52.4

next matches saturday 7 march

i missed the odds this week on the winner which was 2-1 in the liverpool game - that was the only correct prediction again.
i'm going to say it was a scratch week so the total will stay the same and i will pick it up again next round

total +52.4

I am resurrecting this as it popped into my head today that, hold up a minute, 50pts in a short period of time was pretty good, why didn't I continue this?

2 years has passed, xG is an even bigger thing now, have markets adapted? or was i just on a lucky streak

let's pick it up again today by trying to predict today's premier league results and backing in the Correct Score markets

thinking about this now pragmatically it could have just been a lucky winning streak, and there's no real edge, however I do think the Premier League of all leagues in the world may just be the best league to do this on as the results are somewhat predictable. let's see. This is more out of curiosity than anything

the model uses xG averages per home/away game, and I do use rounding up, so if a team has an xG of 1.6 then the scoreline will say they will score 2, so it is optimistically biased, however it has worked out in the past

Aston Villa 1-1 Brentford @ 7.6
Leeds 2-1 Fulham @ 10.5
Southampton 1-2 Arsenal @ 9.4
Wolves 2-1 Leicester @ 11.5
Spurs 2-1 Newcastle @ 10.5

I can see for example that Leeds' xG has been pushed up but mainly by Fulham's high xG away but this is contextual as Fulham have perhaps played some hard games away, giving Leeds an xG of 1.75 - they may not achieve this and it could be skewed, and as the data is only going on last 5 games it could be less accurate. Arsenal's xG today using the model is 1.53 and again that's been rounded up to 2, they could fall short of that and actually only get 1 goal .... interesting nonetheless

2-1 CS has been statistically one of the most probable scorelines in the Premier League over the last 5 seasons

Wolves is actually the most skewed xG today. their home season xG is 0.95 per game but due to Leicester's apalling away record they've been bumped up to 1.54 for today's game. that gives them a predicted score of 2.00. not sure about that and could be a flaw in the weighting of the model (and the rounding up) but there we go. A case could be made for also backing 1-1 in that game....

I've just quickly knocked up some predictions for the French league today, really interesting, and surprisingly high scoring
this league has climbed up the average goals per game over the last few seasons, could be the Messi effect

Angers 2-2 Rennes @ 16.5
Clermont 2-1 Brest @ 10.5
Reims 2-1 Auxerre @ 9.8
Toulouse 2-2 Strasbourg @ 15
Troyes 1-2 Lorient @ 12.5
Nice 2-1 Nantes @ 10
Lille 1-2 Monaco  @ 12

the 2-2 scorelines are particularly interesting, again could be a slight flaw with the rounding up but we can expect goals from these games.

2 correct scores so far

2 correct scores so far

17.8pts profit in France, went mental

England is a blank this GW. today is WHAM 2-1 Bournemouth.

Having thought about this overnight, I think the good run was just that, a good run, and long term whether there's an edge I am doubtful of.  however I will keep it going and see what happens